Resume Les peridotites protogranulaires remontees par les basanites miocenes du Forez representent un manteau lithospherique initialement equilibre dans le facies a grenat, a une profondeur de 50 a 60 km. Elles ont subi une reequilibration dans le facies a spinelle, temoignant d'un rechauffement isobare jusque vers 1 100–1 200 °C. A ce processus est associe un enrichissement en elements a grand rayon ionique, resultant de la percolation de faibles fractions de liquides riches en volatils, voire carbonates. Ces deux processus sont attribues aux effets thermiques et chimiques de l'erosion lithospherique en bordure du panache mantellique du Massif central.
Several studies have reported trace-element data showing that mantle xenoliths from world-wide localities are selectively enriched in uranium (±lead, ±strontium) relative to other incompatible trace elements; these samples show prominent U spikes on primitive mantle-normalized diagrams (e.g., Jochum et al., 1989; Ionov et al., 1995). Most often, the authors have considered these enrichments as “secondary” features related to post-eruption contamination, or weathering. At first sight, this hypothesis is supported by the existence of U spikes in xenoliths that are otherwise devoid of any trace of modal or cryptic mantle metasomatism. Except for the U positive anomaly (±Pb, ±Sr), these samples have a trace-element signature typical of depleted MORB mantle (DMM).
Our ICP-MS study of 31 peridotites representative of mantle xenoliths from Massif Central (France) confirms the widespread character of U enrichment in these rocks. 25 xenoliths have (U/Th)N (= primitive mantle-normalized ratio) and (U/La)N > 1, among which 7 have (U/Th)N in the range 10 - 100 and (U/La)N in the range 5 - 20. However, our results are at variance with previous works in suggesting a “primary” (mantle) origin for the U spike. Evidence is twofold: (1) trace-element variations in whole rocks indicate that the U spike is overprinted by cryptic mantle metasomatism - i.e., overall enrichment of large ion lithophile elements (LILEs) - and (2) acid-leached minerals are enriched in U and account for whole-rock budget. Olivine analyses indicate that fluid-derived inclusions trapped in minerals are a significant host for U. Together with experimental data, this suggests that U enrichment is related to the migration of volatile-rich fluids/melts through the upper mantle. Positive correlations between U/Th, Pb/Ce and Sr/Ce indicate that this process was also responsible for Pb and Sr enrichments.
Two hypothesis may be envisaged for the origin of U-, Pb- and Sr-enriched fluids/melts. They could either be derived from subducted oceanic lithosphere (e.g., during the Variscan orogeny), or they may represent volatile-rich small melt fractions, residual after incomplete solidification of asthenospheric melts at the base of the lithosphere (in the Tertiary). The latter alternative is strongly supported by the existence of negative U anomalies in high-temperature (³ 1200°C) xenoliths considered to represent the base of the lithospheric mantle eroded by uprising asthenosphere (Xu et al., 1998). In this scheme, the development of the U spike in lithospheric peridotites does not require an U-enriched source. As shown with a numerical simulation, it would simply result from selective transport of U (±Pb, ±Sr) by volatile-rich, uprising melts.
A significant implication of this study is that the lithospheric mantle beneath Massif Central is characterized, on overall, by much higher U/Th ratios than both primitive Earth’s mantle and present-day convective mantle. Judging from the number of published data reporting U spikes in mantle xenoliths from various localities, this may also be true for large volumes of subcontinental lithosphere. If timeintegrated, the trace-element signature of lithospheric peridotites may significantly contribute to the isotopic heterogeneity of the mantle.
Evidence for a major heating event accompanied by decompression was recently reported from crustal rocks drilled in the Alboran basin. The metamorphic evolution recorded by these rocks implies complete removal of lithospheric mantle during the Cenozoic, a process that is confirmed by geophysical modelling indicating thin lithosphere beneath the Alboran domain. In this region, the Ronda lherzolite massif (Betic Cordillera, southern Spain) provides a unique opportunity for the observation of mantle processes associated with lithospheric thinning. A striking feature of the Ronda peridotite is a narrow recrystallization front, which has been ascribed to kilometre-scale porous melt flow through the massif. The front separates the spinel tectonite domain, interpreted as old, veined lithospheric mantle, from the granular domain where the lithospheric microstructures, mineralogical assemblages and geochemical signatures were obliterated by grain growth coeval with pervasive infiltration of basaltic melts. On the basis of trace-element abundances in peridotites collected over a distance of 12 km along the recrystallization front, our study confirms that the front is a relatively sharp (≤400 m) geochemical discontinuity at the scale of the Ronda massif. Compared with the spinel tectonites, the coarse-granular peridotites are more homogeneous, more refractory in terms of major elements and more depleted in incompatible trace elements. These features are consistent with a process involving partial melting, kilometre-scale migration of melts by diffuse porous flow and limited melt extraction (2·5–6·5%). Hence, the Ronda recrystallization front is interpreted as the narrow boundary of a partial-melting domain (the coarse-granular peridotites) formed at the expense of subcontinental lithospheric mantle (the spinel tectonites). The existence of melt-consuming reactions in the transitional peridotites, a few hundred metres ahead of the melting front, demonstrates that the front was thermally controlled. This implies that a smooth thermal gradient existed across the Ronda massif during the development of the recrystallization front. Differences in pyroxene compositions on either side of the front may be explained by a transient heating event at ≥1200°C (∼1·5 GPa) coeval with partial melting. Consistent with the geodynamic scenario proposed for the Alboran domain during the Cenozoic, the evolution of the Ronda recrystallization front is considered as an example of thermal erosion and partial melting of lithospheric mantle above upwelling asthenosphere.
Orogenic peridotites display a range of structural, petrological and geochemical characteristics which encompass the deformation styles and compositions expected for subcontinental lithospheric mantle and shallow asthenospheric mantle. In the past, the ambiguous character of these rocks led to somewhat conflicting views on their origin. Recent studies tend to converge on the idea that the orogenic peridotites represent thinned and/or rejuvenated lithospheric mantle, thermally eroded by asthenospheric mantle during continental rifting or early oceanic (or back-arc) initiation. Lithospheric rejuvenation is marked by annealing of deformation microstructures, overgrowth of mineral grains and km-scale modal/chemical variations resulting from partial melt redistribution. Melt-related processes vary between two extremes: partial melting of - and melt extraction from - heated, mildly fertile, lithospheric mantle and refertilization of refractory lithospheric mantle (harzburgites) by asthenospheric partial melts. In the Ronda and Lherz orogenic peridotites, all microstructural, modal and geochemical changes related to lithospheric rejuvenation occur across narrow fronts (² 100 m) interpreted as melt fronts.
The large (~ 300 km2) Ronda massif (Betic Cordilleras, S. Spain) is characterized by a clear-cut ‘recrystallization front’ which can be followed over a distance of ~ 20 km in the massif (Van der Wal & Vissers, 1996 – J. Petrol. 37). Across the front, all the structures of the lithospheric ‘spinel tectonite domain’ were overprinted: the strongly foliated spinel tectonites were converted into ‘coarse-granular peridotites’ while garnet-bearing pyroxenite veins recrystallized as garnet-free websterites. Modal and chemical variations across the front indicate that it was the narrow boundary of a partial melting domain formed at the expense of subcontinental lithospheric mantle. A few hundred meters ahead of the recrystallization/melting front, the presence of a more diffuse and irregular fertilization front (attested by the precipitation of secondary pyroxene in the spinel tectonites) demonstrates that at least a smooth thermal gradient existed across the Ronda massif during the inferred melting event. Therefore, the development and shape of the recrystallization front were very likely thermally controlled. Differences in pyroxene compositions on either side of the front suggest a rapid, transient heating event at ~1200°C and ~ 1.5 GPa. Behind the front, the strongly heterogeneous ‘layered granular subdomain’ records melt redistribution and freezing while the melting domain was receding upon thermal relaxation. In this domain, secondary spinel lherzolites and layered spinel websterites formed by refertilization reactions are intermingled with anastomosed dunite bands interpreted as drainage channels for the last, evolved melt fractions.
The sharpness of geochemical and textural variations occurring at the Ronda recrystallization front indicates coupling between melting and mineral coarsening processes during asthenosphere-lithosphere interaction. Given the importance of melt fraction in the rheology of peridotites, it is likely that such a front may induce a significant drop in viscosity, thus constituting a potential mechanical boundary for the removal of lithosphere by the underlying, convective asthenosphere. The thermal event recorded by the Ronda recrystallization front occurred in a late stage of the massif evolution, in the Cenozoic ; it was aborted by its final emplacement into the crust along extensional detachments. It is thus logical to ascribe this event to the geodynamic evolution of the Alboran Sea, where petrological and geophysical information indicate complete removal of the lithospheric mantle beneath the centre of the basin. The information conveyed by the Ronda massif suggests that positive feedback relationships between lithospheric heating, pervasive melting and convective erosion played a major role in the erosion and removal of the Alboran lithosphere.
The comparatively small (~ 1 km2) Lherz massif (W. Pyrenees, S. France) is made of interlayered spinel lherzolites and websterites in the lower part and highly refractory harzburgites atop the peridotite body. The spinel lherzolites and websterites have a depleted Nd-Sr isotopic signature comparable to that of the MORB-type asthenospheric mantle. In contrast, the harzburgites have enriched isotopic signatures indicating their derivation from LREE- and LILEenriched refractory mantle isolated in the subcontinental lithosphere for at least 1 By - a feature which is further corroborated by Os isotope studies. These two units are separated by a narrow (< 10 m) but extremely convoluted boundary. In map, the complicated shape of the contact results in the isolation of rounded bodies and lenses of harzburgites within anastomosed bands of the spinel lherzolite /websterite suite. However, the continuity of the harzburgite deformation microstructures between individual bodies indicates that they are probably connected to the main harzburgite unit in 3D. Websterite layering and foliation in the spinel lherzolite/websterite suite tend to parallel the contact and are clearly oblique on the pervasive and homogeneously oriented foliation of the harzburgites.
Structural observations and geochemical data indicate that the spinel lherzolite/websterite suite was developped at the expense of the harzbugites via a near-solidus refertilization reaction involving precipitation of pyroxene, spinel and minor amphibole at the expense of olivine and infiltrated melt. The boundary between the spinel lherzolite/websterite and harzburgite units is therefore considered as a reaction front that also traces the upper limit of silicate melt infiltration in the lithospheric mantle. The abrupt modal and geochemical variations occuring across the front are coupled with significant textural changes involving annealing of the harzburgite deformation microstructures, mineral grain coarsening and the development of new deformation structures and microstructures parallel to the front - as exemplified in the field by isoclinal folding of websterite layers with axial planes parallel to the contact with harzburgites. Geochemical evidence for the refertilization reaction includes (1) transient (chromatographic), convex-upward normalized REE patterns in cpx at the harzburgite-lherzolite transition, (2) nearly constant concentration of moderately incompatible elements in the lherzolite minerals (e.g., Ti in cpx) and (3) anomalously high Ni content in olivine of the lherzolite near the contact with harzburgite.
Based on the geochemical signature of the spinel lherzolite/ websterite suite, the infiltrated melt responsible for harzburgite refertilization in Lherz was probably generated by decompression melting of upwelling asthenospheric mantle beneath heated/thinned lithosphere. However, all mineral compositions were reequilibrated at much lower temperatures (~ 800°C) than the near-solidus conditions implied by this event. This indicates either that the lithospheric mantle underwent slow thermal relaxation or/and that a significant span of time has elpased between the thermal relaxation and the crustal emplacement of the peridotite body in the Cretaceous. The most likely episode to place the formation of the Lherz refertilization front is therefore the late- Variscan, post-collisional thermal event responsible for granulitic metamorphism in W. Pyrenees.
The Ronda and Lherz melting/reaction fronts display major differences as well as striking resemblances. Ronda shows a clear-cut, thermally controlled, melting front reflecting the existence of a partially molten domain developed at the expense of lithospheric mantle. In contrast, Lherz is characterized by a convoluted reaction front formed by the coalescence of reactive melt infiltration channels filled by asthenospheric melts. In part, these differences may simply result from the significant differences in lithosphere composition in the two localities. Although heterogeneous on a cm to tens of meters scale, the lithospheric mantle preserved in the Ronda massif (the ‘spinel tectonite domain’ of Van der Wal & Vissers, 1996) is fairly fertile on average, due to the predominance of lherzolites over harzburgites and the presence of mafic layers (garnet pyroxenites). For such a composition, substantial degree of partial melting would occur when the lithospheric mantle is conductively heated above the peridotite solidus. In contrast, the lithospheric mantle preserved in the Lherz massif (the upper harzburgite unit) is strongly homogeneous and refractory, being virtually devoid of clinopyroxene and amphibole. For such a composition, even heating well above the peridotite solidus will not trigger partial melting.
However, the observations in Lherz indicate that conductive heating of refractory lithospheric peridotites may allow the infiltration of asthenospheric partial melts, leading to the ‘thermo-chemical erosion’ of the lithospheric mantle. The microstructural variations observed at the Lherz reaction front are comparable to those observed at the Ronda melting front, suggesting that the two mechanisms have similar implications in terms of mantle rheology and lithospheric mechanical erosion. In particular, the existence in the refertilized domain of a high-temperature deformation parallel to the front demonstrates that the Lherz reaction front was a sharp mechanical boundary for convective lithospheric erosion.
In the Forez district, two volcanic events are documented. The earlier one, of Paleocene age, predates the Oligocene rifting. The magmas (melilitite) are of carbonatitic affinity and they would correspond to small volume melts responsible for the chemical and mineralogical modifications previously identified in the lithospheric mantle of this region. The later lavas are Lower Miocene and of basanitic composition. Their isotopic signatures (Sr, Nd, Pb) suggest that they partly originated from partial melting of this modified lithosphere. The Paleocene magmatism would indicate the initiation of asthenospheric upwelling beneath the French Massif Central. Deux événements volcaniques ont été identifiés dans le Forez. Le premier, paléocène, est antérieur au rifting oligocène. Les magmas de nature mélilititique pourraient correspondre aux liquides riches en éléments incompatibles et en volatils responsables des modifications chimiques et minéralogiques identifiées dans le manteau lithosphérique de cette région. Le second événement, du Miocène inférieur, est post-rifting et basanitique. Les compositions isotopiques suggèrent que ces basanites proviennent en partie de la fusion partielle de la lithosphère modifiée par la percolation d'un fluide ayant la signature de la mélilitite. Le magmatisme paléocène marquerait le début de la remontée asthénosphérique sous le Massif central.
A well-know characteristic of the Ronda peridotite massif (S Spain) is the existence of a km-scale structural (Van der Wal and Vissers, 1996), petrological (Obata, 1980) and geochemical (Frey at al., 1985; Reisberg et al., 1989) zoning. From N-NW to the S-SE, three structural domains of decreasing ages have been recognized by Van der Wal and Vissers (1996): (1) a spinel-tectonite domain made of porphyroclastic spinel peridotites, garnet-spinel peridotites and subordinate garnet pyroxenites, representing old, veined lithospheric mantle (Reisberg et al., 1989, 1991); (2) a granular peridotite domain composed of coarse-grain spinel peridotites and subordinate spinel websterites, both derived from spinel tectonites via open-system recrystallization at shallow depth (Van der Wal and Bodinier, 1996; Garrido and Bodinier, 1999); (3) a plagioclase tectonite domain, the youngest of the three domains, composed of porphyroclastic plagioclase peridotites related to crustal emplacement of the massif (Van der Wal and Vissers, 1996). Compared with the petrological zoning identified by Obata (1980), these three domains correspond respectively to the (1) garnet lherzolite facies and ariegite subfacies, (2) Seiland subfacies and (3) plagioclase lherzolite facies.
One of the most striking features of the massif is the transition from the spinel tectonite to the coarse-grained peridotite domain. This transition occurs across a narrow recrystallization front (£ 400 m) where major microtextural, traceelement and isotopic variations take place in peridotites and pyroxenites (Van der Wal and Bodinier, 1996; Garrido and Bodinier, 1999). The recrystallization front is roughly vertical and can be followed over a distance of more than 20 km. It has been interpreted by Van der Wal and Bodinier (1996) as a magmatic permeability barrier for basaltic melts accumulated in the granular domain. At regional scale, the development of the front was probably connected with a recent thermal event in the Alboran domain, ascribed to upwelling of asthenospheric mantle after gravitational collapse of thickened continental lithosphere. Hence, the Ronda recrystallization front provides a unique opportunity to study asthenophere-lithosphere interactions associated with mantle upwelling.
The aim of this study was twofold: (1) evaluate the lateral continuity of geochemical variations across the front, and (2) investigate whether the recrystallization was associated with a thermal event registered by mineral compositions. 40 peridotites collected along several sections across the front were analyzed by ICP-MS for trace elements. Orthopyroxene and clinopyroxene pairs were analyzed by electron microprobe. In addition, these minerals were separated from 6 samples and analyzed by ICP-AES to estimate equilibrium temperatures while avoiding compositional variations produced by pyroxene exsolution.
Most of the analyzed peridotites display LREE-depleted normalized patterns. Yet, a systematic difference is observed for CeN/SmN (= chondrite-normalized Ce/Yb ratio) across the recrystallization front, the granular lherzolites being systematically more LREE depleted (CeN/SmN < 0.5) than equivalent rock types in the spinel tectonite domain. These data confirm that the front represents a major geochemical discontinuity in the Ronda massif and was developped in open system. Likewise, the equilibrium temperatures measured with the two-pyroxene thermometer of Brey and Kholer (1990) are significantly higher in the granular peridotites (1190-1220°C) than in the spinel tectonites (1070-1090°C).
The higher temperature obtained for the granular peridotites suggests that the driving force of recrystallization was thermal in origin. In addition, independant field and geochemical constraints on melt processes indicate that a significant temperature gradient existed across the Ronda massif during the development of the recrystallization front. In the granular domain, evidence for partial melting and pervasive interaction with basaltic melts implies temperatures above or close to the peridotite solidus. About in the meantime, the spinel tectonites were traversed by low-temperature, volatile-rich small melt fractions similar to those inferred for diffuse cryptic metasomatism in subcontinental lithosphere (Bedini et al., 1997). The outermost parts of the spinel tectonite domain were even cut by intrusive dikes (Garrido and Bodinier, 1999).
However, considering that the recrystallization front is < 400 m wide, the temperature difference measured in our samples (100 - 125°C) would imply an unreallistically steep geothermal gradient (ca. 200°C / km). Hence, the observed difference is more likely of kinetic origin. Since diffusion in solids is orders of magnitude slower than diffusion in melts, the higher melt-rock ratios inferred for the granular peridotites, together with the dissolution-crystallization process, probably led to faster mineral reequilibration in this domain, compared with the spinel tectonites. In order to preserve a fossil recrystallization front and differences in equilibrium temperatures across the front, the thermal event must have been relatively short-lived. The kinetic interpretation requires that the time scale of heating was intermediate between the critical time of diffusion in solids and that of melt-enhanced inter-granular diffusion and recrystallization. Since the thermal event occured in a very late stage of the massif evolution, it has probably aborted due to fast crustal emplacement of the massif by extensional tectonics.
A corollary of the above discussion is that thermomecanical and magmatic processes are probably coupled during lithosphere-asthenosphere interaction. This relationship has not being considered in pure thermo-mechanical models of lithospheric erosion. Yet, feedback relationships between melt accumulation and olivine grain growth can cause a substantial drop in the viscosity of peridotites and create a km-scale weak layer at the base of the lithosphere. Mechanical erosion of such layer by the convective mantle will trigger a new cycle of thermal and chemical erosion of the lithosphere.